This is only a preview of the July 1993 issue of Silicon Chip. You can view 37 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Articles in this series:
Articles in this series:
Items relevant to "Build A Single Chip Message Recorder":
Items relevant to "Light Beam Relay Extender":
Items relevant to "Build An AM Radio Trainer; Pt.2":
Articles in this series:
Items relevant to "A Low-Cost Quiz Game Adjudicator":
Articles in this series:
Articles in this series:
|
SERVICEMAN'S LOG
When it looks easy, it often ain’t
Yes, it did look easy. There it was; an obviously
damaged component clearly visible. All I had to
do was find out why it was damaged. Although
this would involve some searching, it turned out
to be a much bigger search than anyone could
have imagined.
One of the most elementary methods of servicing has always been
visual observation. Way back in the
very early days of radio, when bright
emitter valves were the norm, the
first thing one looked for in a dead
set was whether all the valves were
alight. In fact, there were those who
bemoaned the advent of the dull
emitter valves, significantly more
economical though they were, because they no longer provided this
visual clue.
Much has changed since then of
course, but the visual clue remains
a valuable one, even with today’s
technology. The burn marks on a PC
board, the bulging capacitor, the blackened fuse, the burnt resistor; they all
pinpoint a fault area. And while they
don’t necessarily pinpoint the fault
itself, they can show one where to
start looking.
All of which is leading up to a
particularly frustrating problem I
encountered recently; the more so
because at first glance – literally –
there was a typical visual clue which
should have put me straight on the
right track.
It all started when the customer
turned up with a Samsung colour set,
model CB-5012Z. This is a 51cm set
using a P/58SC type chassis and was
about three years old. The complaint
was straightforward enough; the set
was completely dead, having simply
failed in the middle of a program.
So, at the first opportunity, I put
it up on the bench. I didn’t bother to
switch it on but simply pulled the back
off and looked for any obvious clues.
This was relatively easy because all the
parts are on a single PC board, the only
reservation being that the components
in the power supply section are very
tightly packed.
Two things were immediately obvious: (1) the mains fuse, F801 (3.5A),
was blown; and (2) resistor R809
(270kΩ, 1W) was badly blackened.
This resistor runs from the positive
side of the bridge rectifier, at about
300V, to pin 4 of the switchmode power supply control IC, IC801 (TDA4601).
And, as I subsequently dis
covered,
safety resistor R801 (5.6Ω 7W) in the
mains input line, just after the fuse,
had also been sacrificed.
cial about all that. Fairly obviously,
there was a short that involved all
three components and, as such, I
didn’t think that it would be hard
to find. Anyway, the first thing to do
was to replace the faulty resistors and
I did this without even testing them.
This done, I replaced the fuse and
switched on with everything under
close scrutiny.
Splat! There was a flash of flame, a
puff of black smoke, and I had another
blackened resistor. Fortunately, a fast
reflex action by my switch finger saved
the fuse and the safety resistor.
My initial thought was simply along
the lines that a short at the IC end of
R809 could produce such symptoms. I
even went so far as to check for a short
circuit between pin 4 of the IC and
chassis; it was almost a reflex action.
But then, on reflection, I realised that
this didn’t make sense. Even putting
a 270kΩ resistor directly across 300V
would dissipate only about one third
of a watt.
So what was destroying the resistor?
All I could think of was that a much
higher voltage, from somewhere else
in the power supply, was finding its
way to this resistor. But from where
and by what means remained a mystery.
I went over the circuit around pin
4 of IC801 and resistor R809 but drew
a complete blank. Finally, and somewhat against my better judgement, I
decided that it must be a faulty IC. In
any case, replacing it would prove the
point, one way or the other.
The only snag was that I didn’t have
this particular IC in stock, so one had
to be ordered. When it arrived a couple
of days later, I lost no time in fitting it.
This proved to be a somewhat tricky
exercise due to the rather cramped
conditions on this part of the board
and the fact that the IC is mounted on
a heatsink.
Splat No.1
Splat No.2
Well, there was nothing very spe40 Silicon Chip
Eventually, the job was completed
Fig.1: the power supply circuit for the Samsung CB-5012Z. Fuse F801 is at
extreme left, safety resistor R801 to the right, & the bridge rectifier to the
right again. R809 is below the lower left corner of IC801 at top right, while
C816 is mid-way up the right-hand edge of the diagram.
and I made ready for another test. A
new 270kΩ resistor had been fitted and
I hoped all would go well this time.
I pressed the power switch. Splat!
Another flash of flame, another puff of
smoke, and another blackened resistor.
I gave up!
Well, almost but I certainly felt like
it. Unfortunately, I had no choice but
to keep at it and so, for want of a better
ap
proach, I simply began checking
every component around the IC, either
measuring then in-situ or removing
them from the board for testing where
necessary.
I had checked a dozen or more components in this way, without result,
and was beginning to question the
wisdom of this approach when I found
myself in the vicinity of transistor
Q801, the power supply switching
transistor. This was removed and tested but also proved to be OK.
The next component was C816, a
222pF 1000V ceramic capacitor connected between Q801’s collector and
chassis. This component is obviously
a spike suppressing device. Because
of the associated circuitry around it,
I decided that this it would also have
to be removed for testing.
In fact, pulling it out was all the testing needed. It was mounted so close
to other components that I could see
only one side of it. But when I pulled it
out and the other side became visible, I
realised that I had struck oil. The case
had split open to reveal a great black
gaping crack.
So at last I’d found the real culprit.
But what, you may ask, did it have to
do with resistor R809, which appears
to be in no way connected with this
part of the circuit. And if you are thinking of way-out explanations involving
spikes in Q801’s collector circuit, forget it. Maybe there were some spikes
but that isn’t the explanation.
In fact, it was much more mundane
than that and simply hinges on the
proximity of C816 to R809. They were
sitting side by side, virtually touching,
with C816 lying slightly over the top
of R809.
So the smoke and flame I had
observed had come from C816, not
R809. And the blackening of R809?
This was almost certainly a burn – not
from internal heat but from external
heat generated by C816. Remember, I
mentioned earlier that I had not even
bothered to check the “damaged” resistors. That was a fatal mistake.
Had I done so, I would almost
certainly have adopted a dif
ferent
approach. When I eventually checked
all three of these resistors, they were
spot on in value. There was nothing
wrong with any of them. The damage
was purely cosmetic and I had been
well and truly conned.
Rubbing in the salt
But there was still some salt to be
rubbed into the wound. I replaced
C816, fitted a new resistor for R809
purely for appearance, and switched
on. And up came a perfect picture; the
only thing that had ever been wrong
with the set was C816. And it had
carried a perfect visual clue but one
which was impossible to see.
Had I been able to see it, I would
have simply replaced the capacitor
and the set would have been back in
operation in a matter of minutes. As
it was, I wasted hours on the job and,
financially, it was a total disaster;
something which had to be written
off to experience.
In that sense, it wasn’t a complete
loss. Apart from the obvious lessons,
one other point emerged. I realised
that there was a failure pattern
emerging concerning the C816 type
capacitor.
Quite recently, I had also serviced a
couple of Samsung chassis which carried the Akai label. Both suffered from
the same fault – failure of a capacitor
across the horizontal output transistor.
And it was an identical capacitor: blue
ceramic, 222pF, 1000V. In both cases,
July 1993 41
SERVICEMAN'S LOG – CTD
But my customer knew where; onto
the power mains connected to his TV
set. After that, the TV set didn’t go any
more. I wonder if the Greek gods know
about TV sets?
OK, enough! But it was classic case
of a mains lightning strike and this
customer wasn’t the only one affected.
When I pulled the back off the set, the
damage was plain to see.
The most obvious was the mains
fuse, F801, 4A. The inside of the glass
was totally blackened, suggesting a
pretty violent strike, and I had no
doubt that I would find more subtle
damage as I went along.
The other visual clue involved a line
filter, L801, in the mains lead immediately following the fuse. The filter
coils themselves were undamaged but
the white plastic case which enclosed
them had been blown to pieces. Since
it was still working, I decided to leave
it until later.
The fuse was replaced and I moved
on to the next item down the line:
the bridge rectifier involving diodes
D801-804. Two of these four diodes
had snuffed it and these were replaced.
Switch-on
they had simply developed a dead
short and shut the set down without
any fireworks. But from now on, I’m
keeping my eye out for any faults
which might involve this particular
capacitor. It could well be less reliable
than one expects from this type in
general service.
Further to that observation, I have
been able to secure another make of
capacitor which I hope will be less
troublesome than the originals. When
I needed replacement capacitors for
the Akai sets, it was more convenient
to order them from an independent
supplier rather than from Samsung.
These not only carry a different brand
but, more importantly, are rated at
2000V.
These new capacitors were fitted to
the Akai sets, as well as to the Samsung set which was the subject of this
month’s story. Here’s hoping that I
have struck a blow for my customers.
hardly the set’s fault. No, the blame
really lies with the great god Jupiter.
In a fit of pique, “he hurled a thunderbolt into the air, which fell to earth
he knew not where” (as they say in
the classics).
Jupiter strikes
Fig.2: parts layout for the power
supply in the Samsung CB-5012Z.
R809 (circled) is situated between
transistor Q801 on the right & C816
on the left. Note that, in practice, the
board layout is much more crowded
than this diagram indicates.
My next story is also about a Sam
sung set – a model CB-518F fitted
with a P50HA chassis – and it also
in
volves visual clues. But I must
hasten to add that this problem was
42 Silicon Chip
OK, time for a switch-on test. This
left no doubt that there was more
trouble ahead. There were loud
protestations from the switchmode
section of the power supply, suggesting a serious overload. I immediately
checked the HT rail for any suggestion of a short to chassis but could
find nothing wrong.
On this basis, and because all the
faults so far had been at the input to
the power supply, it seemed likely
that the fault was still in this area.
There are several more diodes in this
section and I checked all these but
found nothing wrong.
The next suspect was the regulator
IC, Q801 (STR50103A). But before
taking a final step in this direction, I
made a few more checks. I was able
to measure some HT rail voltage –
about 68V as compared to the 103V
shown on the circuit (pin 2 IC Q801
and TP103) – and I also checked the
horizontal output transistor (Q404,
2SD-1555) but this appeared to be OK.
At that stage, I felt that I had gone as
far afield as was reasonable for a strike
of this kind, so I returned to the regu
lator IC. This is a small device, having
only five pins, and I had one in stock
so it was a simple matter to replace it.
ing circuit, caused the two coils to be
pushed apart slightly, due to magnetic
repulsion. They sat only loosely on the
ferrite core.
The movement wasn’t very great,
and would not have caused any
damage in the normal way. But with
the massive surge that destroyed the
bridge diodes and the fuse, the movement had obviously been much great
er; enough to break the flimsy plastic
cover. So that solved that particular
mystery.
Unanswered questions
This photo clearly shows the crack in the back of C816. Also shown in the
blackened fuse (F801) and one of the replacement resistors used for R809.
But all that did was establish that
there was nothing wrong with the
original IC; the new one made no
difference.
I made a few more checks and found
that the 12V rail was down in about
the same proportion as the HT rail
loss. This 12V rail is derived from a
16.5V tap (pin 2) on the horizontal
output transformer via diode D408
and resistor R225 (47Ω, 2W).
But there is also a 12V rail derived
from the chopper transformer via diode D820, which is used as a starting
supply to get the horizontal system
running. And at this point I couldn’t
be sure which of these two supplies
was powering the system, to the extent
that it was working at all.
I also realised that, while all this
analysis of the circuit was very interesting, it wasn’t really revealing
anything that might help solve the
problem. It was time to change tactics.
let-down after all the chasing around
the circuit.
And what about the line filter I
mentioned earlier? While the coils
were undamaged in any way, the white
plastic cover was scattered in pieces
around the inside of the cabinet. How
come?
I found the answer quite by chance.
The filter consists of two fine wire
coils (or chokes) wound on small flat
plastic bobbins, about the diameter of
a 5-cent piece, but somewhat thicker.
These in turn are mounted side by
side on the centre leg of a rectangular
ferrite core.
And I noticed, when switching the
set on after it was repaired, that the
switch-on surge, due to the degauss-
But that still leaves other questions
unanswered. Why did such a massive
surge, having destroyed the bridge
diodes in that part of the circuit, skip
over the regulator IC and the horizontal
driver stage, to pick on the horizontal
output stage? And why didn’t it spread
further via the supply rails and do a
lot more damage?
More importantly, from a practical
point of view, why did Q404 test OK
when it wasn’t?
I don’t have any answers for the
first two questions. I doubt whether
anybody has – except Jupiter perhaps
and he’s not telling.
I don’t have a complete answer to
the third question eith
er, but there
seems little doubt that the protective
devices in these transistors (ie, the
diode between collector and emitter
and the resistor between base and
emitter) make them difficult to test
reliably. The resistor, in particular
(normally 35-40Ω), makes it difficult
Transistors cheat
Speculating on likely component
failures, my thoughts came back to the
horizontal output tran
sistor, Q804.
Granted, I had run the meter over it
and decided that it was OK. But it
wouldn’t be the first time that such
a transistor had cheated the testing
procedure.
As always, and as they used to say
in the old valve days, the ultimate test
of a suspect device is to replace it.
Which was what I did, it not being a
particularly difficult procedure.
And that was it. The set was up and
running in all its original glory. Which
was both a relief and something of a
July 1993 43
SERVICEMAN'S LOG – CTD
Fig.3: the power supply circuit for the Samsung CB-518F. The mains on/off
switch is at the bottom left of the diagram. Fuse F801 follows, then the line
filter L801, the degauss circuit L802, and bridge rectifier D801-804. Next in
line is chopper transformer T801 then and switching IC Q801. The horizontal
deflection circuit is at the top of the diagram.
to determine the condition of the
base-emitter junction.
On the other hand, they don’t always confuse the issue; sometimes
faults are quite readily detected. It all
depends on the nature of the failure.
So the rule seems to be if it tests faulty,
then it is faulty; if it tests OK, it might
44 Silicon Chip
be faulty, or it might not.
Must try not to get caught like that
again.
Circuit diagrams
One final comment. Unfortunately,
the quality of the diagrams in many
manuals leaves a lot to be desired,
and the diagrams for the Samsung sets
just discussed fall into this category.
The main problem stems from the
large size needed for many original
drawings, followed by over reduction
in an effort to accommodate them in a
typical manual.
This can create major problems
when trying to trace a circuit, while
tracking down a difficult fault. Component values are often hard to read,
particularly where figures 6, 8, 9 and
even 0 (zero) are concerned. In a blurr
ed reproduction, one can easily be
mistaken for the other.
Even more confusion can occur
where circuit lines cross. While the
concept of using small circular blob
to denote a con
nection, or no blob
to denote a non-connective crossing,
has the advantage of draughting simplicity, it falls down badly where the
reproduction is poor.
A certain amount of image spread
can occur where lines cross, creating
the impression of a blob where none
exists, or giving rise to doubts as to
whether a genuine blob is really only
a blur. And take my word for it; it can
waste a lot of time.
Personally, I much prefer the more
conservative drawing convention,
which uses a loop to denote a non-connective crossing. The stated objection,
of course, is that this requires more
work and is therefore more costly.
Well maybe it used to be but these
days, with Computer Aided Drawing
(CAD) programs, I doubt whether the
difference is all that great.
Anyway, for my money, the differSC
ence is worth any extra cost.
|